Troubleshooting Jump Sensor Overreads on Soft Surfaces: Mats, Turf, and the Contact-Delay Fix

A physio you work with sends over the numbers from a post-op knee assessment: 34 cm on the force plate in the clinic gym, 41 cm on the padded landing mat you use pitch-side for the same athlete a week later. Nothing about the athlete's presentation supports a 20% jump between two sessions. Coaches see the same pattern when a countermovement jump test moves from a wood gym floor to a turf field for pre-season screening, or when a return-to-sport check gets shifted onto a thick crash mat for landing safety. The sensor did not get worse and the athlete did not suddenly get better. The surface changed how long it takes for a foot landing to register as ground contact, and that handful of extra milliseconds is enough to make an unremarkable jump look like a breakthrough.

This is not a rare edge case. Any flight-time-based jump sensor, including inertial units that infer height from time spent in the air rather than measuring force directly, is exposed to it, and the error gets worse rather than better for exactly the athletes you are most likely to test on a padded surface: post-surgical patients doing early landings, younger athletes, and anyone producing a shorter, lower-effort jump. What follows is the mechanism behind the overread, the research it traces back to, surface-by-surface numbers you can act on, and a calibration protocol that lets you keep testing on the mat you actually own without quietly inflating every number that comes off it.

Why a Soft Surface Makes a Jump Sensor Read High

An accelerometer-based jump sensor, PoinT GO's IMU included, never measures height directly. It measures the time between takeoff and landing and back-calculates height from projectile motion: h = g·t²/8, where t is flight time and g is 9.81 m/s². Takeoff is flagged the instant vertical acceleration collapses toward free fall as the foot leaves the ground. Landing is flagged the instant a deceleration spike crosses a fixed threshold, typically in the 2–3g range for PoinT GO's default jump algorithm. On a rigid floor that spike is close to instantaneous, because the foot decelerates almost the moment it touches down, so threshold-crossing happens within a couple of milliseconds of true ground contact.

A compliant surface changes the shape of that curve entirely. Some of the athlete's downward momentum goes into compressing the mat instead of decelerating the body, so the deceleration signal rises gradually and takes longer to cross the same fixed threshold. Ferris, Louie, and Farley (1998) documented the underlying mechanic in a study of hopping and running across surfaces of varying stiffness, showing that as surface compliance increases, effective ground contact stretches out because the surface itself absorbs and returns energy over a longer window rather than simply flattening the impact. Their apparatus was a mechanical hopping and running platform, not a jump sensor on a gym mat, so it does not hand over a ready-made correction factor for foam or turf – but it establishes exactly why any threshold-based contact detector reads late on a surface with give, whether that give comes from a laboratory rig or a folded crash mat on your training floor.

The Quadratic Trap: Why a Small Delay Becomes a Big Number

Because flight time is squared in that formula, a timing error does not translate one-to-one into a height error. A contact-detection delay on top of true flight time inflates the recorded height by roughly double its share of that flight time, which means the same delay barely dents a long, powerful jump and meaningfully distorts a short one.

Take a 40 cm countermovement jump, which corresponds to a true flight time of about 0.571 seconds. A 25 ms landing-detection delay pushes the recorded flight time to 0.596 seconds, which back-calculates to 43.6 cm – a 3.6 cm, roughly 9%, overread. Run that identical 25 ms delay through a 20 cm jump, a true flight time of about 0.404 seconds, and the recorded height climbs to 22.6 cm: only 2.6 cm in absolute terms, but a 13% overread, meaningfully worse in relative terms for a jump barely half the height. Attia, Dhahbi, Chaouachi, Padulo, Wong, and Chamari (2017) reported a comparable magnitude of flight-time bias using photocell timing gates against a force-platform reference, with countermovement jump height overestimated by an average of 4.7 ± 2.3 cm, roughly 18%, once the flight-time formula's sensitivity to takeoff and landing detection was accounted for. Their setup used photocells over a standard rigid floor, not a compliant mat, so it is not a direct surface study – but it is independent confirmation that flight-time methods carry meaningful, systematic bias before surface compliance is ever added on top.

That asymmetry matters clinically. The athletes most likely to be tested on a padded surface – post-operative return-to-sport patients doing early box or mat landings, younger or lower-capacity athletes, anyone working through a deliberately low-load session – are also the athletes producing the shortest jumps, which is exactly where this error bites hardest.

Which Surfaces Actually Distort the Reading

Nuzzo, Anning, and Scharfenberg (2011) compared a contact-mat jump system against a force platform and found systematically higher jump-height readings from the mat device, tracing the gap to the mat's binary switch-closure method for flagging takeoff and landing rather than to any single surface property. That is the same class of vulnerability a compliant surface adds to, not a separate problem: any device that infers ground contact from a threshold event, whether a mat switch or an accelerometer spike, is exposed to a longer registration delay once the surface underneath it has give. Their comparison ran on a single indoor floor with a limited sample of 17 participants, so it does not quantify a mat-versus-turf gradient either, which is the gap the field calibration protocol below is built to close at your own facility.

The table compiles that mechanism against PoinT GO's multi-site field calibration testing, comparing simultaneous 240 fps video ground-truth flight time against sensor-reported flight time across the surface categories partner facilities actually use. Treat these as planning ranges to confirm on your own floor, not as a substitute for the calibration walkthrough in the next section.

Field Calibration: Measuring Your Own Surface's Delay

The table above tells you whether to worry. This protocol tells you exactly how much to correct for, on the specific mat or turf you actually have.

1. Pick a reference surface. Assign one rigid, sensor-friendly surface at your facility – a gym floor, a force plate, or thin rubber flooring – as the fixed baseline every athlete gets measured against at least once.
2. Film in parallel. Set a phone or action camera to 240 fps, framed side-on at the foot-ground interface, for every calibration trial on the target surface.
3. Run five max-effort countermovement jumps on the target soft surface with the IMU recording as normal.
4. Extract ground-truth flight time from video. Step frame by frame to the last frame with visible ground contact before takeoff and the first frame of visible ground contact on landing; at 240 fps each frame equals roughly 4.17 ms.
5. Compute the delta. Subtract the sensor-reported flight time from the video-derived ground-truth flight time for each of the five trials and average the result – that average is your surface's contact-registration delay in milliseconds.
6. Save it as a surface profile. Enter the measured offset into a named surface profile in the PoinT GO app for that mat or turf area, so the app subtracts it before computing height on any future session logged against that profile.
7. Recheck quarterly, or sooner after a mat swap. Foam compresses and turf infill settles with use, so a profile calibrated in preseason can drift by 5–10 ms by mid-season; a fresh five-trial check takes under ten minutes and keeps the offset honest.

Standardizing the Protocol Across Sites and Testers

A calibration offset only protects you if every tester actually uses it the same way, every time.

• Log the surface profile with every session, not just the jump number, so a review six months later can tell which correction, if any, was applied.
• Never compare raw jump height across two different surface profiles. Compare corrected values, or better, keep longitudinal tracking data on a single athlete tied to a single surface profile whenever possible.
• Treat an uncalibrated one-off session as descriptive only. If an athlete gets tested away from home – a borrowed facility, a tournament warm-up area – flag that session and hold off on using it for a training-load or return-to-sport decision until it is corroborated on the calibrated home surface.
• Standardize equipment across sites where you can. The same mat brand and thickness at every location means one calibration profile travels with the equipment instead of needing to be rebuilt at each site.

Common Mistakes That Make This Worse

• Blaming the sensor and recalibrating the device electronics instead of the surface underneath the athlete. The IMU itself is not the source of the drift; the ground it is landing on is.
• Comparing a rehab session on a crash mat directly against a baseline captured on a hard floor months earlier, without ever applying an offset to either number.
• Assuming a thicker mat is always worse. Thickness matters less than compression stiffness – a dense 3 cm competition landing mat can behave closer to the semi-compliant category than a soft, loosely packed 6 cm practice mat.
• Averaging jumps taken across mixed surfaces within a single session as though the numbers sat on the same scale, which quietly blends two different bias levels into one misleading mean.